Brynjolfsson and McAfee (2014) argue that the transformations brought about by digital technology will be profoundly beneficial. The new era will be different, but also better due to an increase in both the variety and the volume of our consumption. West (2011) points out other positive outcomes, including progress in education (distance learning), healthcare (sharing digital images over the internet), energy efficiency (smart grids), communications (video conference). Many of the applications of new technology by government, as noted above, have a beneficial effect.
On the other hand, new technology raises question about social and individual values, privacy, security, and societal impacts. There will be unpleasant consequences of the new technology that must be managed. We have attempted to identify the major implications of the new technology or the questions that are raised by the new technology.
The following are implications/effects of the adoption of new technology. In some cases, we are uncertain about the consequences of the new technology, so we can only raise the issue. Generally, the implications addressed in this section are what would generally be considered “negative” effects. The following list should not be considered comprehensive.
One potentially significant effect of new technologies will be on the labor market, i.e., employment. Furman and Seamans (2018) outline a broad framework for thinking about how AI and its applications might affect labor markets. First is a theoretical perspective; second, the historical and empirical approach to the effects of technological change; third, making predictions of how these new technologies will impact labor markets in the near future.
Examined from the theoretical prospective, Furman and Seamans suggest four effects on labor markets:
- Innovation can directly replace human labor in the affected sector.
- Automation can create new jobs in existing sectors or new sectors. For instance, Mandel (2017) finds that new jobs created by e-commerce in fulfillment and call centers made up for the lost jobs in traditional brick-and-mortar retail stores.
- New AI technology will boost labor productivity and increase total income in the economy. Higher income will increase demand for all jobs, including jobs in non-technology sectors. Furman and Seamans suggest that growth in the leisure and hospitality sectors is an example of this process in the United States.
- Technology will alter the way workers perform certain tasks, automating routine tasks and freeing up workers to specialize in the higher-level functions that cannot be automated.
The next two frameworks—the historical and empirical approach to the effects of technological change and making predictions of how these new technologies will impact labor markets in the near future—can be summed up as trying to determine whether the experience for labor with AI will be different than for previous technological advances.
In a comprehensive essay summarizing past effects of new technology on labor markets, Autor (2015), emphasizes that technology acts as a complement as well as a substitute for human labor. (The history of automatic bank teller machines and human tellers is an instructive example.) The complementary role of new technologies, including AI, is often neglected in current thinking. Autor suggests that the adjustment of labor markets to AI will be similar to past experiences with new technology. However, the breadth of sectors and workers affected will be larger.
Furman and Seamans (2018) also suggest the current AI development path is similar to past experiences labor has had in adjusting to new technology. They do not foresee AI replacing a broad range of unique human abilities and skills. However, if AI were to develop more quickly and be able to work across various sectors and skill sets, becoming a “general AI,” then the disruption to the labor markets could be greater.
Given current levels of AI and forecasted improvements, several authors and organizations have tried to predict what the near future holds for the labor market in terms of sector and skill specific changes. The following offer a sample of short descriptions of the various findings in the literature.
Frey and Osborne (2017) find that 47 percent of U.S. employment is at high risk of automation.
A report examining developed countries by Arntz et al. (2016), in contrast, finds that roughly 9 percent of jobs in the U.S. and across OECD countries will be highly susceptible to automation.
The McKinsey Global Institute (MGI) in a 2017 report (Bughin, et al. 2017) estimates that at least 30 percent of activities are automatable in about 60 percent of occupations. But, the MGI Report also cautions that such automation will not necessarily substitute for labor and reports that less than a fifth of respondents said AI was being adopted to reduce labor costs. Rather, respondents report that AI is used to improve capital efficiency or enhance existing products. A similar study done by the World Economic Forum (2018) finds that 50 percent of responding firms expected new automation would lead to some reduction in their full-time workforce by 2022. In addition, 54 percent of all employees will need significant improvements to their existing skills by 2022, due to changes in technology.
MGI in another 2017 study, “A Future That Works: Automation, Employment, and Productivity” (Manyika et al. 2017), estimates skills, occupations, and sectors subject to automation. It also tries to estimate how quickly new technology featuring AI and robots might be implemented.
The literature suggests that there is a strong relationship between the occupations or skills that can be automated and income or education. The Council of Economic Advisors (2016) used the Frey and Osbourne characterizations and found that jobs making less than $20 per hour had an 83 percent probability of automation. Jobs making over $40 per hour only had a 4 percent probability of automation. Arntz et al. (2016) find that in developed OECD countries, jobs that require a high school degree or less are much more likely to be automatable than jobs with a college or graduate degree.
Furman and Seamans (2018) offer several broad concerns regarding the effects AI will have on labor. First, part of the reason that AI does not displace labor will be because relative wages adjust, adding to the existing problem of wage inequality. Second, the pressure on lower-skilled jobs may contribute to the continuing decline in labor force participation for prime-age workers. Finally, while AI is most likely to boost productivity, it has the potential for large disruptions to the workforce. Thus, it will be important to ensure that appropriate policies exist to protect workers, such as ample opportunities for retraining and perhaps some protections to basic income while these adjustments take place.
As technology permeates more and more aspects of daily living, some groups of adults are increasingly left behind because they have less access and fewer skills than others (Robinson et al., 2015). Research shows both the disproportionate disadvantages and potential opportunities that technology holds for certain marginalized groups, including women (Horrigan 2016), African Americans (Prieger, 2015), people who have been incarcerated (Western, Braga, Davis, and Sirois 2015), people with disabilities (Andersen and Perrin 2017; Dobransky and Hargittai 2016), people who are LBGT, people living in poverty (Andersen 2017; Horrigan 2016), and people in rural communities (Horrigan 2016). Below we provide a few examples of key implications of technology, both positive and negative, for these marginalized communities.
Even though women’s educational attainment now exceeds that of their male counterparts (Ryan and Bauman 2015), gender disparities have worsened as a result of new technology and a changing economic landscape (Shire 2015). One contributing factor is that women lag behind men with regard to technology use. Compared to men, women use technology less often, for shorter durations, and perceive that they possess fewer skills (Robinson et al. 2015).
Poverty is a common roadblock to internet access, use, and skills. African Americans who are more likely to live in poverty have fewer options for broadband internet (Prieger 2015). African Americans have identified two key deterrents to accessing broadband: the first is the cost of broadband services, and the second is that broadband companies rely on credit scores to determine eligibility for services (Perrin 2017). As a result, African Americans tend to rely on mobile phones more for internet access (Perrin 2017; Prieger 2015; Ryan 2017). A downside of smartphone use during job searches is that job applicants struggle to be able to read job postings, to upload necessary documents, or input all of the text required for applications (Smith 2015). But, smartphones have also helped African Americans increase their civic engagement (e.g., researching political information, connecting with government resources, and reading news) (Mossberger, Tolbert, and Anderson 2017).
People who are incarcerated are disproportionately poor and African American. In addition, technology can change rapidly during a person’s incarceration (Western et al. 2015). Unfortunately, when technology is available to inmates it is on closed systems that are closely monitored; some systems only allow inmates to receive (not send) messages, and inmates can be charged as much as $1.25 per email message (Sobol 2018). On the other hand, when people who are incarcerated have access to technology, it can provide them opportunities to gain job skills, to stay socially connected to their support system, and to ensure a successful reentry back into society upon release.
Because people with disabilities tend to be poorer and older, this population also has lower rates of internet use. Of those individuals with disabilities who use the internet, they have lower engagement in online activities than their counterparts without disabilities (Dobransky and Hargittai 2016). Like other groups, however, people with disabilities could use technology to increase their socialization and to gain access to employment markets.
The gig economy, which has increased in size with the growth of technological advancements, also affects marginalized groups differentially. LGBT (Mejia and Parker 2018), African Americans (Edelman, Luca, and Svirsky 2017), and women (Barzilay and Ben-David 2016) experience discrimination in the gig economy. Barzilay and Ben-David (2016) examined gender disparities on one digital platform that connected jobseekers with specific tasks that needed to be completed. Barzilay and Ben-David (2016) found that women earn on average one-third less than men, even after controlling for ratings, occupation, and hours worked. These communities can be affected as consumers in the gig economy as well. For example, LGBT customers experience higher cancelations on ride-sharing platforms (Mejia and Parker 2018). These groups also experience discrimination as consumers in platforms like Airbnb. People with African American-sounding names (Edelman et al. 2017; Leong and Belzer 2016) and people in same-sex relationships (Ahuja and Lyons 2017) have lower acceptance rates as Airbnb guests. Likewise, African Americans wait longer for Uber and Lyft rides (Calo and Rosenblat 2017) and pay more for purchases on eBay auctions (Leong and Belzer 2016).
Despite the critiques of the gig economy, it also provides some opportunities for marginalized groups to access services and seek employment that otherwise may not exist.
New technology could change market structures and the sizes of businesses. For example:
- Electronic shopping platforms provide sellers access to detailed information regarding consumer shopping patterns. This could lead to price discrimination (Goolsbee 2018).
- It is not clear what the economies of scale are for e-businesses. If there are large startup costs but small costs to expand the network of users, then it is expected that such firms will be large, leading to monopolistic power and higher prices.
- Similarly, the need for large data sets for AI is a likely barrier to entry. This means that market structures will be concentrated, leading to higher prices. That raises issues of how to regulate such quasi-monopolies (Furmans and Seamans 2018).
- On the other hand, new technology lowers the cost of information, which could reduce barriers to entry in a number of industries, and the cost of transactions.
- Information and communications technologies give access to almost any kind of information, making information cheap and ubiquitous (Stehn 2002). The result is an increase in the information content of goods and in the information intensity of production processes.
New technology can be both a cause of crime and a tool in the prevention of crime and law enforcement. There are implications for security, crime, and justice. The concepts of privacy, security, crime, and justice change as we move to a cashless society, increase the use of facial recognition, etc. How do we deal with the rise in cybercrime, identify theft, ransomware, and misleading algorithms (Pinguelo and Muller 2011)? Jin (forthcoming) provides a lengthy discussion of the size and consequences of consumer privacy issues that arise from new technology.
Cybercrime—any crime that somehow involves digital technology—encompasses many different types of crimes that takes place in the online environment. Our highly digitized society presents many opportunities for people to: use computers to engage in crime, focus on computers as the object of crime, study crime, and respond to it.
The scope of cybercrime and the efforts to respond to it, use it to enhance public safety, and study how cybercriminals behave are vast and wide. One account estimates cybercrime costs are expected to double to $6 trillion by 2021. Other estimates by the Center for Strategic and International Studies and McAfee (Lewis 2018) put the loss at about 0.8 percent of global GDP or $600 billion a year currently, up from $445 billion in 2014. These big numbers, although derived from different methodologies and with big differences between them, point to a common and inescapable conclusion: cybercrime’s impact on the global economy is large and likely to increase. They also provide motivation for governments and businesses to respond to the large and growing impacts of cybercrime.
Digital technology is being used to study cybercrime and improve public safety.
David Maimon and his colleagues (Testa et al. 2017; Howell et al. 2017) have advanced the idea that technology is key to developing theoretical insights on the development and progression of both online and offline criminal events. Technological tools, in this sense, are also key for the deployment of research initiatives that support the collection of data on both online and offline crime, particularly for the purposes of understanding the unfolding of a criminal event (i.e. offender decision makers). For more details, see Maimon and Louderback (2019).
In the context of online crime, cybersecurity experts have devoted considerable attention in the last 20 years to developing tools and policies designed to detect computer and network vulnerabilities and prevent these crimes from developing. Developments in AI and machine learning and the creation of highly focused machine learning algorithms that can perform the key tasks of automated detection and deception of manual attackers, if properly designed, could be deployed to detect cyber intrusions.
On the traditional crimes front, AI applications are being developed for the analysis of video and image data, DNA testing, gun shots, license plate detection, and crime forecasting. The advantages of these machine-based tools are that they are consistent in the mistakes they make and they can process lots of data quickly. For example, the Atlanta Journal-Constitution (Capelouto 2018) recently reported on license plate reading technology that was used to identify a person with outstanding warrants who was subsequently arrested. On the downside, questions have been raised about big data methods used to assist criminal justice decision makers (such as in pretrial detention or parole release) and whether the algorithms have built-in racial or ethnic biases that stem from actors outside the data used to compute scores or the manner in which the algorithms were trained to develop their predictions.
Overall, AI and big data in the digital world provide opportunities to enhance public safety, but if used uncritically can perpetuate long-standing patterns of unequal treatment of racial minorities.
- Will new technology result in changes in power structures?
- What are the effects on trust and cooperation? What happens to the need for trust?
- Will job loss results in a rise of populism? See Levy (2018) for a discussion.
Trajtenberg (2018) discusses what he calls “democratization of expectations,” namely, that individuals in advanced economies have come to expect full participation politically, economically, and culturally. He worries that if there is a sharp split between the winners and losers from the adoption of new technology, it could have serious consequences that could threaten the very fabric of democracy. Francois (2018) makes an equivalent point. If computers replace humans, and if income goes to the owners of capital, there will be “human unrest on a massive scale.” He suggests that to prevent these consequences, we will have to find a way to share the returns to capital and to “create meaning from non-work related activities.”
New technology has reduced the important of distance or space, given the increased ability to communicate remotely. Electronic platforms such as M-Tuck eliminate the need for face-to-face contact between employer and employees. To the extent that robots replace humans in production, the relative importance of other production inputs, such as land and energy, should change. These and other changes brought about by changing technology could have significant consequences for where businesses and workers will locate and where non-workers will choose to live. This, in turn, could change the nature of urbanization. Residential patterns could change, the need for urbanization could increase or decrease, and the relative importance of rural versus urban areas could be affected.
Advances in technology (computer/genetics) is expected to have major impacts on health and healthcare. For example, genetically targeted cancer treatments are being developed, individual and population data are guiding treatment patterns, and communications between patients and providers and among providers are changing. See Jiang et al (2017), Health IT Analytics (2018) and Gray (2018) for discussion and examples. New health information technology (HIT) is being used to develop health information systems that are expected to, among other things, increase efficiency and reduce errors.
On the other hand, it is not clear what the effect of the general adoption of new technology on health will be. As was mentioned, autonomous vehicles will reduce traffic accidents. There will be implications for retirement and pensions if new technology such as AVs results in longer lives due to reduction in traffic accidents.
New technology could have significant effects on developing countries, perhaps changing developing countries’ comparative advantage, and lead to a significant global economic structural change.
- Bremmer (2018) and Rodrik (2018) raise the following concerns regarding developing countries.
- Governments and institutions in developing countries aren’t ready for the impact of new technology.
- Technological change will sharply reduce the low-wage advantage that helps poor countries and poor people climb the economic ladder.
- Wealthy countries have access to education systems that prepare citizens to adapt to new economic and social realities. Developing countries do not.
As a result, the net impact of new technology on developing countries looks considerably uncertain.
- New technology could affect international trade. Brynjolfsson, Hui and Liu (2018), for example, find that machine translation systems on eBay have increased international trade on that platform by 17 percent.
Return to the implications for Government and Non-Profit Management.
Return to the Implications section of the “Identifying the Landscape of New Technology” report.
Proceed to the implications for Public Policy and related fields.